Study Of Short-term Load Forecasting Based On Random Forest Regression Algorithm

Load forecasting is the foundation and premise of a safe operation in power system,and it is also the key for the power sector to arrange the schedule and power supply plan.Based on the market-oriented operation of the power sector and changing need of power load,the short term load forecasting performance has higher requirements,thus this paper introduces the Random Forest Regression(RFR)and Support Vector Machine(SVM)algorithms which have a superior performance in recent years for the short-term load forecasting.Many other scholars have different opinions about their performance comparison,my own view is put forward based on the case analysis.Firstly,this article introduced the research meaning and purpose about the short-term load forecasting,and then simply introduced some of the commonly used algorithm principle and the advantages and disadvantages in the present situation of the research.Secondly,according to the New York City power load information,the factors and characteristics of load are analyzed,and according to the periodic law of power load which is applied in the actual forecast.Newly,the origin of random forest and the principle of the algorithm are described in detail,and the preparatory work of case analysis is made on the analysis of characteristic variables' importance and correlation.Lastly,with the New York City's electric power load and weather data,the RFR algorithm were applied in the field of short-term load forecast,and the RFR and SVM algorithm were comprehensively analyzed,then concluded that: RFR has higher prediction accuracy and better generalization performance in short-term load forecasting;the performance superiority of RFR and SVM cannot be generalized,that is the specific examples has the specific analysis.Through the study of this paper,we have a detailed understanding of the short-term load forecasting based on the random forest regression algorithm,and it has played a fundamental role after in the comparison between RFR and SVM prediction algorithm.